CHRISTOPH SCHULZ, NEAR COLOGNE
I use these tools every day, and I write the explanations I wish I had found when I started.
I did not always get AI. A few years ago the words machine learning made my eyes glaze over like everyone else’s. Then I started actually using the tools, and I found the gap between what they can do and what most people know about them is enormous, and almost entirely unnecessary. Closing that gap is the whole point of this site.
Counted on 6 September 2026. The tool and language numbers are the ones behind the API cost calculator.
The stack I actually pay for
Not a recommendation list. This is what is on my own invoice.
Most author pages tell you what someone thinks about AI. That is cheap to write. Here is the harder and more useful thing: what I actually run, what I use each one for, and where each one annoys me.
| What I pay for | What I use it for | My honest take |
|---|---|---|
| Claude Max (the 5x plan) | Opus 5 for research and anything fact-heavy. Opus 4.8 for long running threads. Sonnet or Haiku when I just need help with phrasing. | The one I would not give up. Picking the smaller model for small jobs is most of what keeps me inside my quota. |
| Gemini Pro | A second opinion when I want to see whether an answer holds up twice. | The answers come back too simple too often, so I end up asking again and again. Steering it is harder, and it loses the thread more than I would like. |
| ChatGPT | The free tier, occasionally. | Where I started. Not where I stayed. |
| Adobe (with Firefly) | The credits go into the image and video generator, without a watermark on the output. | The watermark-free output is genuinely the reason the subscription stays. |
I have switched models more than once, and I expect to switch again. Anyone who tells you there is one right model is selling something.
On a flat plan you are not buying dollars. You are buying headroom. Spend it on the jobs that need the big model, and stop spending it on the ones that do not.
Where I learned this
Anonymised on purpose. More on that below.
| When | What | What it taught me |
|---|---|---|
| Today | Business expert, global claims transformation, at a financial services company | What actually happens when a large organisation tries to put AI into a real process rather than into a slide. I also co-run one customer-facing process across the German and Australian organisations, which is where I learned that the same process means two different things in two countries. |
| 2024 to 2025 | Management trainee, insurance, at a global financial services group | How a regulated business really works underneath the org chart. This is also where I built my first AI applications: internal training tools, and process improvements in individual areas. Not pilots for a steering committee. Things colleagues actually used. |
| Before that | Store manager at a national grocery retailer | My first job where the numbers and the people were both mine. A hard school. |
| Studies | Business, with a focus on marketing and innovation research | My thesis was on organisational error reporting: how people and systems learn from mistakes. Dry on paper. It has stayed with me. |
Why no company names. This site is a private project. It is not connected to my employer, it does not speak for my employer, and nothing here is written on their time or with their tools. Naming them would blur that line, so I do not.
What is on the page instead is the part that actually transfers: I sit where strategy meets implementation and I watch what works and what quietly does not. That is what shapes almost everything I write here. I am not theorising about AI from the outside.
What this site is, and what it is not
So you know within ten seconds whether to stay
What you will find
- Interactive tools that answer a question with your numbers, not mine
- Step-by-step guides to specific AI tools
- Plain-language explanations of the concepts you actually need
- Honest assessments of what these tools are good at, and what they are not
What you will not
- Breaking news. Covering every release is a full-time job and other people do it better
- Hype. AI is a significant technology, not a religion
- Doom. It is also not the end of the world
- Jargon for its own sake, and no condescension either
I am neither an evangelist nor a sceptic. Both positions strike me as intellectually lazy. Most of the day-to-day reality is far less dramatic than either camp suggests: a set of tools that are sometimes very useful, sometimes overhyped, and always changing.
How I use AI to write about AI
The disclosure, because you should not have to guess
I use AI tools in my writing process. It would be strange not to, given what this site is about. Research, first drafts, editing.
Everything published here is reviewed, edited and signed off by me. The ideas, the judgments and the editorial choices are mine. Where an article contains a number, I checked it against the primary source and I say when I checked it.
I think that is a reasonable and honest way to use these tools. If you disagree, I am interested in why. Contact is below.
Get in touch
A question, a topic you want covered, or feedback on something I got wrong: I would like to hear it. The contact page is the fastest route.
And if something here was useful, send it to someone who would also find it useful. That is the whole point.
Christoph Schulz. Near Cologne, Germany. Writing at LearnAI24 since November 2024.